Author: Godot
Storage oligopoly benefits and operating leverage surge under HBM cycles
Viewing Micron as a storage company is the biggest misconception in its valuation.
The market still perceives Micron through the lens of a traditional cyclical stock, severely underestimating the fundamental differences between this AI-driven cycle and historical cycles.
This article will analyze Micron’s profit revaluation logic from three dimensions: industry physical structure, oligopolistic competition discipline, and pricing mechanisms.
The fundamental difference between DRAM and NAND: capital expenditure efficiency determines the depth of the moat
Viewing Micron as a storage company is the biggest misconception in its valuation.
DRAM and NAND differ fundamentally in physical structure, technological barriers, and capital return rates.
The core storage unit of DRAM is a capacitor, used to store charge. As the process advances from 1-alpha to 1-beta and now to the current 1-gamma node, capacitor sizes continue to shrink, making leakage issues increasingly challenging and forcing manufacturers to adopt EUV lithography machines, each costing hundreds of millions of dollars.
This means that even if second-tier manufacturers spend heavily, they still struggle to catch up to the process advantages of leading players.
NAND is completely different. 3D NAND increases capacity by vertically stacking memory cells, and the industry is currently moving from 232 layers to over 300 layers.
Although this technology path also consumes capital, the barrier is relatively manageable. Secondary manufacturers such as Kioxia and Western Digital can relatively easily increase supply by investing capital to purchase stacking equipment.
This is precisely why NAND's gross margin has historically been lower than DRAM's and more susceptible to price wars.
Micron's current revenue mix consists of approximately 70% DRAM and 30% NAND. Under a SOTP valuation, the DRAM business should command a higher valuation multiple, while the NAND business should be assigned a discount.
The core logic behind buying Micron is to capitalize on the oligopoly advantages of DRAM and the growth potential of HBM.
Oligopoly: The True Anchor for Valuation
The storage industry has undergone a brutal consolidation, reducing dozens of players to just three remaining: Samsung, SK Hynix, and Micron.
The key variable for assessing industry health is not market share competition, but capital expenditure discipline (CapEx Discipline). Once oligopolists aggressively expand production to grab market share, the entire industry suffers losses. Once they默契 control supply, profits grow exponentially.
After enduring the previous deep downturn cycle, the three major players have demonstrated unprecedented restraint on the supply side. The more critical variable is HBM’s consumption of capacity, which constitutes the largest structural difference between this cycle and previous ones.
HBM capacity squeeze effect: Underestimated supply-side shortage
The market widely recognizes that HBM is expensive and offers high margins, but underestimates the extent to which HBM is physically crowding out traditional DRAM capacity.
This mechanism is composed of two factors combined:
1) Die Size Penalty: HBM consists of multiple layers of DRAM dies (typically 8 or 12 layers for HBM3E) stacked atop a single base logic die. To achieve ultra-high bandwidth, significant routing space must be reserved within the chip. For the same capacity, an HBM occupies approximately 2 to 2.5 times the silicon area of traditional DDR5.
2) Yield Loss: HBM relies on through-silicon vias to penetrate multiple layers of dies and connect them with copper pillars. Any misalignment or failure in drilling through any layer results in the entire HBM being scrapped, causing HBM’s overall yield to be significantly lower than that of standard DRAM.
Two factors combined mean that producing one wafer of HBM consumes the capacity that could otherwise produce 3 to 4 wafers of standard DDR5. Even if demand recovery in PCs and smartphones remains slow, as long as the three major players allocate their best capacity to NVIDIA for HBM production, the supply of traditional DRAM will be artificially constrained, thereby supporting and pushing up industry-wide average prices.
Therefore, as long as HBM demand does not experience a cliff-like decline, Micron’s pricing power in traditional DRAM remains extremely strong. The market should not easily predict a recurrence of price wars.
Pricing mechanism reassessment: shifting from bit growth-driven to ASP-driven
Micron's revenue formula is: Revenue = Total shipped bits × Average Selling Price (ASP).
Historically, storage cycle booms have been driven by bit growth and ASP doubling, with typical scenarios occurring during periods of smartphone explosion.
However, the characteristics of this cycle are markedly different: bit growth has remained modest, at low to mid-single digits, while the overall ASP has risen significantly due to the inclusion of HBM. HBM prices are typically several times higher than those of standard DRAM and are locked in through advance payment mechanisms and long-term agreements.
In the past, Micron's performance fluctuated sharply with the spot market, but now a significant portion of its capacity is locked into high-margin, long-cycle AI orders. The peak gross margin of this cycle is expected to surpass historical highs.
Operating leverage: Non-linear profit acceleration under a fixed cost structure
Semiconductor manufacturing is a typical capital-intensive industry with high fixed costs. The construction of an advanced wafer fab can cost between $15 billion and $20 billion, with equipment depreciation typically amortized over 5 to 7 years.
Assume Micron's base cost to produce a batch of wafers is $100, of which $40 to $50 is fixed depreciation expense on equipment that must be recorded regardless of production volume. The remaining variable costs, such as materials and labor, account for approximately half.
When the ASP increases from $100 to $120, the additional $20 is nearly 100% converted into operating profit, since fixed costs remain unchanged. This is the power of operating leverage.
The same logic applies in reverse: during a downturn, if ASP falls below the fixed cost line, companies face severe cash flow losses. This is also why storage stocks exhibit price volatility far exceeding their fundamental fluctuations.
Summary,
When building Micron’s financial model for the next several quarters, assumptions about gross margin expansion should be more aggressive than in traditional cyclical models. HBM locks in high-margin capacity, oligopolistic discipline suppresses price competition, and operating leverage amplifies profit elasticity from rising ASPs.
Micron's valuation represents a structural reassessment of its profitability. This is the ideal window for institutional capital to build positions, as the market still prices Micron using the traditional price-to-book framework for cyclical stocks.
Of course, the core risk variable that requires continuous monitoring is HBM end-demand. Once cloud providers slow their capital expenditures or NVIDIA GPU shipment trends show a turning point, the foundation of the entire logic chain will weaken.
CapEx, technological pace, and HBM packaging bottlenecks
CapEx博弈推动股价上涨
In most industries, management cutting capital expenditures is interpreted as a loss of confidence in the future. But storage is the opposite.
Due to the industry's strong cyclical nature, the ratio of capital expenditure to revenue is a key indicator for determining whether supply is out of control.
Health levels are typically around 30%; once they surpass 40%, it signals that the entire industry has entered a phase of frantic production expansion, inevitably leading to overcapacity and a bloody price war in the future.
During the super downturn cycle from 2022 to 2023, Micron's management decisively cut its wafer fab equipment spending for fiscal years 2023 and 2024 by nearly 50%.
Wall Street has assigned a high premium; reducing CapEx has not only safeguarded the bottom line of free cash flow but also sent a clear signal to the market about price support.
When professional analysts see this type of announcement, they do not lower their target prices; instead, they view it as the first buy signal indicating a cycle bottom and a potential price reversal.
Additionally, CapEx can be split into two components: factory facilities and wafer manufacturing equipment.
There is a significant time difference between these two types of expenses. Building a semiconductor fab takes 2 to 3 years, while moving and commissioning equipment such as lithography machines takes only 6 to 12 months.
Micron reduced its equipment spending by 50% during the downturn, directly lowering short-term output and supporting market prices; however, capital expenditures for facility construction, such as initial investments in new plants in Idaho and New York, were not significantly cut.
This strategy avoids disrupting supply and triggering a price war in the short term, while reserving physical space for the next AI supercycle.
Once a demand inflection point emerges, Micron can quickly procure equipment to fill its facilities and capture market share. This is a classic counter-cyclical investment mindset.
Micron technology roadmap
Precise timing of 1-gamma nodes
The technical roadmap is essentially an ROI issue. An ASML EUV lithography machine can cost hundreds of millions of dollars and consumes enormous amounts of electricity.
If the cost savings from the technological upgrade cannot offset the massive depreciation of new equipment, it will be financially devastating.
Micron has taken a path distinctly different from Samsung over the past few years. At the 1-alpha and 1-beta nodes, Micron pushed traditional DUV technology to its limits, achieving mass production earlier than Samsung, while strategically delaying the full adoption of EUV.
As a result, Micron's manufacturing costs on these nodes are lower than Samsung's, which adopted EUV first, giving Micron a lead in capital efficiency over its peers.
At this point, entering the current 1-gamma node, the physical dimensions have become so small that the cost of DUV multi-patterning exceeds that of direct EUV usage, at which time Micron began fully and maturely integrating EUV equipment—landing precisely at the optimal intersection of technological cost reduction and equipment depreciation.
In our financial model, we expect Micron’s 1-gamma node yield ramp to be smoother than Samsung’s initial EUV introduction, directly translating to improved gross margin performance.
HBM3E's overtaking maneuver
In the AI era, whether a storage vendor is NVIDIA-certified determines its pricing power.
The HBM market has long been dominated by SK Hynix, while Micron skipped HBM3 and redirected all its R&D resources toward the more advanced HBM3E.
Successfully integrated into NVIDIA's H200 and B100 supply chain, with power consumption nearly 30% lower than competitors.
In data center scenarios, power consumption directly determines cooling costs and is a core metric for hyperscaler selection. As a result, Micron’s HBM capacity for 2024 and 2025 has been fully pre-ordered.
This change has fundamentally impacted the valuation methodology. The HBM business is no longer suitable for valuation using the traditional P/B framework for cyclical stocks, but should instead be valued using a P/E multiple for growth stocks within an SOTP approach. This is the core driver of Micron’s valuation reconstruction.
Bottleneck: HBM isn't manufactured—it's assembled
The market widely knows that HBM is in short supply, but the real production bottleneck is not in front-end silicon manufacturing, but in back-end advanced packaging.
The core HBM process involves stacking eight or twelve ultra-thin DRAM dies and connecting them with tens of thousands of tiny through-silicon vias and copper pillars.
Any misalignment or thermal issue at any layer can render the entire expensive HBM chip unusable, making HBM yield and capacity expansion highly dependent on the maturity of the packaging process.
More critically, there is a parasitic relationship in the supply chain. Micron’s HBM3E cannot be delivered directly to end customers; it must be sent to TSMC to be integrated with NVIDIA’s GPUs on a single packaging substrate via CoWoS.
This means Micron’s HBM revenue growth trajectory is tightly aligned with TSMC’s CoWoS capacity expansion curve.
Evaluating Micron's HBM revenue requires more than just monitoring Micron's own production capacity plans—it must also closely track TSMC's CoWoS capacity expansion rate.
If TSMC’s packaging process stalls, even the largest volume of HBM produced by Micron cannot be converted into revenue for the quarter. This is one of the most critical cross-verification metrics that institutional analysts track across the entire AI computing supply chain.
Summary,
Dram oligarchs maintained price floors through tacit coordination, preserved cash flow with counter-cyclical CapEx strategies, enhanced capital efficiency with precisely timed EUV roadmaps, and unlocked valuation ceilings by betting on HBM3E generation leap.
Focus on four key high-frequency metrics: the split ratio between WFE and Shell in Micron’s quarterly CapEx, the yield ramp progress of the 1-gamma node, changes in NVIDIA’s supply chain allocation of Micron’s HBM share, and TSMC’s monthly CoWoS capacity data.
Any turning point in these four indicators could trigger a reassessment of the target price. Among them, TSMC's CoWoS capacity is the most easily overlooked yet most direct factor determining Micron's near-term revenue realization and requires close monitoring.
Micron's cyclical investment indicator system
It’s often the peak time to sell when a company releases its most impressive earnings report with an unbelievably low P/E ratio; conversely, it’s typically the best time to buy when the company reports massive losses and a disastrous income statement.
This counterintuitive phenomenon stems from the fact that stock pricing discounts future fundamentals 6 to 9 months in advance.
Since historical financial statements are not useful, what forward-looking indicators do institutional analysts rely on to identify turning points?
Days Inventory Outstanding (DIO)
For memory chips, inventory is poison.
Chips that sit too long not only depreciate but also tie up significant working capital. The balance sheet metric that institutional analysts monitor most closely is DIO (Days Inventory Outstanding).
Historical data shows that Micron's DIO range typically falls between 90 and 110 days. Once DIO exceeds 130 days and continues to rise, it indicates that the produced chips are not selling and are piling up in inventory.
Even though quarterly profits appear acceptable, a price war is imminent and the rating must be lowered immediately.
But more importantly, watch for bottoming signals.
In the most severe period of 2023, Micron's DIO once rose to over 200 days.
The true buying opportunity is not when DIO returns to a healthy 100-day level, but rather the moment DIO begins to decline环比 from its 220-day high, such as dropping to 180 days.
This means inventory reduction has already begun, and the fundamentals have hit their bottom. By the time DIO returns to normal levels, the main upward trend in the stock price will already be over.
You can't just look at Micron's own warehouse
Once the chips are sold to customers such as Dell, Apple, and Amazon AWS, they are removed from Micron's balance sheet and recognized as revenue.
But this does not mean the chips have been truly consumed; they may have simply been moved from Micron's warehouse to the customer's warehouse.
Institutional analysts build extensive channel tracking models, analyzing the financial reports of PC manufacturers such as Dell, HP, and Lenovo, as well as major cloud service providers, to reverse-engineer the weeks of inventory for components held by customers.
At the beginning of a downturn, end-demand starts to weaken, but customers continue purchasing due to long-term contracts. At this stage, Micron’s financial results still look strong, but customer inventories are already full. Once a major customer suddenly announces a halt in shipments to clear existing stock, Micron’s revenue could instantly halve.
A true bottom signal must satisfy two conditions: Micron's own inventory begins to decline, and channel customer inventory simultaneously drops to normal levels, typically within 4 to 6 weeks.
The true beginning of Micron's primary upward trend occurs when large clients exhaust their inventories and are forced to re-enter the market to initiate restocking cycles.
Is inventory impairment a bearish or bullish signal?
The core is the accounting LCM principle (Lower of Cost or Market Value), and the logic is this: suppose Micron’s cost to produce a chip is $100, but due to industry downturn, the market price drops to $80. Under accounting standards, Micron must immediately recognize a $20 inventory write-down. This causes the gross margin for the quarter to collapse instantly, potentially turning deeply negative.
The book cost basis of this batch of chips has been forcibly lowered to $80. When the market recovers next quarter and prices rise back to $95, Micron will report a $15 profit on the sale of this inventory, not a loss.
When Micron announced a large inventory write-down during its earnings call, retail investors often panic and sell, but institutional investors understand this is management cleaning the books.
After shedding the burden, the gross margin elasticity over the next few quarters will be significantly amplified. This is the accounting foundation for why stock prices rise immediately after negative factors are priced in during the early stage of a cycle reversal.
The income statement is fake; the cash flow statement is real.
Inventory write-downs can lead to a collapse in gross margin and a sharp drop in net profit, but this does not mean Micron will go bankrupt. This knowledge gap is the fundamental reason why short sellers are repeatedly liquidated at the bottom of the cycle.
Inventory impairment is essentially a non-cash expense, representing only an accounting adjustment without any actual cash outflow.
More critically, during this industry downturn, Micron will significantly reduce raw material purchases and halt inventory accumulation, causing changes in accounts payable and inventory to generate positive cash inflows on the cash flow statement—this is the reverse working capital release effect.
During periods of extreme market cold, institutions don't look at Micron's P/E or EPS because they're all negative and meaningless.
Focus closely on free cash flow (FCF). If Micron reports a $2 billion net loss for the quarter but free cash flow remains positive or only slightly negative, it indicates management has halted the bleeding through operational measures.
Cash flow bottoms out 1 to 2 quarters before profit does, making it one of Wall Street’s most profitable leading indicators.
When DRAM and NAND prices plummeted to rock-bottom levels due to oversupply, smartphone manufacturers found that upgrading from the standard 8GB of RAM to 12GB or even 16GB added minimal cost while serving as a key marketing advantage—resulting in an immediate surge in RAM capacity per device.
This automatic doubling of order capacity triggered by a price drop will quickly absorb excess inventory in the market.
When industry chain research shows that mid-to-low-end smartphones and ordinary PCs are beginning to widely standardize large-capacity memory, the bit inflection point has arrived, and the inventory reduction cycle will end at a pace far exceeding expectations.
Q2 FY2026 Earnings Deep Dive: When an 8x P/B Becomes the New Normal
Total quarterly revenue reached $23.86 billion, a nearly 200% increase compared to $8.05 billion in the same period last year.
The DRAM business contributed the largest share, reaching $18.768 billion, a 207% year-over-year increase.
More important than the revenue figures is the internal structure of revenue growth.
The MD&A of the financial report explicitly states that the significant growth in DRAM revenue was primarily driven by a mid-60% sequential increase in average selling price (ASP), while bit shipments saw only low single-digit growth.
This is direct evidence of complete pricing power. Micron doesn’t need to sell more chips to profit—the structural shortage of HBM has given it the ability to raise prices.
SOTP valuation
Summing up the SOTP valuation, among the four business units (BUs) disclosed in Micron's earnings report, data from two core BUs are particularly critical.
1) Cloud storage business
Q2 revenue reached $7.749 billion, nearly tripling compared to $2.947 billion in the same period last year.
Even more alarming is the operating profit margin of 66%.
Micron has gained pricing power over HBM and high-capacity DDR5 for data centers.
2) Mobile and Client Services
Traditionally, memory for mobile phones and PCs has been highly competitive, with gross margins under long-term pressure.
However, the revenue for this business in Q2 reached $7.711 billion, compared to just $2.236 billion in the same period last year, with an operating profit margin reaching an astonishing 76%, up from 47% in the same period last year.
After HBM drained wafer capacity across the entire industry, the capacity spillover effect caused severe structural shortages in memory for ordinary smartphones and PCs, enabling Micron to achieve rare windfall profits in its traditional business.
This is the result of cyclical and structural shortages reinforcing each other.
Gross profit margin breaks through historical highs
Q2 non-GAAP gross margin reached as high as 74.9%, with Q3 guidance pointing to approximately 81%.
This level of gross margin is typically found only in pure software companies like Microsoft or AI chip design companies with absolute monopolies.
It is an unprecedented phenomenon in the storage industry over the past 30 years for a capital-intensive wafer manufacturing company to deliver such high-quality earnings data.
Discrepancy between inventory and accounts receivable
A divergence between inventory and accounts receivable signals a strong seller's market.
Despite a 75% quarter-over-quarter increase in Q2 revenue, ending inventory stood at just $8.267 billion, slightly below the $8.355 billion at the end of the previous fiscal year.
The chips were almost entirely picked up without staying in the warehouse.
Accounts receivable surged from $9.265 billion at the end of the previous fiscal year to $17.314 billion at the end of Q2, nearly doubling.
It's not that customers are defaulting on payments; rather, the rate of restocking and price increases has been too rapid, with giants like NVIDIA and Microsoft rapidly pulling in massive quantities, and enormous payments still within 30- to 60-day credit terms.
This signals an extremely strong positive surge in operating cash flow for the next quarter. Days Inventory Outstanding (DIO) is at a remarkably healthy low, with no signs of oversupply.
Net cash provided by operating activities reached $11.9 billion, net capital expenditures after government subsidies amounted to $5.0 billion, and adjusted free cash flow reached $6.899 billion.
The management also announced a 30% increase in the quarterly dividend, a clear signal of strong confidence in future cash flows.
However, capital expenditures for fiscal year 2026, net of government subsidies, are expected to exceed $25 billion. Micron is aggressively expanding capacity in Idaho, New York, and even at its newly acquired $1.8 billion semiconductor fab in Taiwan.
This means that much of the massive cash generated by Micron will be immediately reinvested in equipment purchases and factory construction. It’s a double-edged sword.
Based on Q2 FY2026 financial data, quarterly revenue of $23.86 billion, Non-GAAP diluted EPS of $12.20, Q3 gross margin guidance of approximately 81%, diluted outstanding shares of approximately 1.142 billion, and current book value per share of approximately $64.20.
Over the next 6 to 12 months, corresponding to the second half of FY2026 through FY2027, the market will be entirely dominated by HBM supply shortages and software-level profit margins.
Wall Street has completely abandoned the traditional P/B model and is now using SOTP sum-of-the-parts and high-growth P/E models to price Micron.
Micron is no longer just a memory chip manufacturer, but a core asset in AI computing infrastructure, benefiting from NVIDIA's valuation premium.
Based on management guidance, Q3 EPS is expected to reach $19.15, with a range of ±$0.40.
Extrapolating this momentum, the annualized EPS could easily exceed $75. This level of profitability is unimaginable in the traditional storage industry.
Under the SOTP framework, apply a forward P/E multiple of 20 to 25 for AI/HBM-related businesses (current actual: 27.22), and a high-end P/B multiple of 2.5 for traditional businesses.
The overall target is derived by summing the two segments.
- Target price: $1,300 to $1,500
- Target market capitalization: $1.48 trillion to $1.71 trillion
This will be the absolute peak in the history of semiconductors.
Retail investors and trend-following CTAs will flood in, but institutional funds will respond by strictly implementing phased selling as the stock price approaches $1,400.
The core judgment here is that when everyone is using the P/E ratio of growth stocks to price cyclical stocks, the valuation anchor has become detached, and it’s only a matter of time before it snaps.
Any hint from any of the four major cloud giants that AI returns are falling short, a slight softening in HBM spot prices, and TSMC's CoWoS capacity expansion outpacing NVIDIA's demand growth.
The appearance of any one of these three signals should trigger a position reduction, and the simultaneous appearance of all three indicates the end of Stage One.